Joint OECD / Eurostat questionnaire on national accounts employment and hours worked ITALY

Transcription

1 Joint OECD / Eurostat questionnaire on national accounts employment and hours worked ITALY I. Employment 1 Question 1.0: Describe succinctly the architecture of your estimation method for employment in persons. Employment estimates result from a labour account run by national accounts department in ISTAT 2. These estimates are also different from LFS or other official employment figures. The calculations are benchmarked in the year 2001, and extrapolated for current years. The basis of the method is the comparison 3 of labour estimates from supply and demand side of the labour market. Labour supply and demand are aligned conceptually and imbalances between them are associated to conceal or secondary jobs, according to certain rules (see below). This comparison supply-demand is possible because of the exceptional availability in Italy of 3 important sources at the same moment of time: Census of Industry, Trade and Services (CIS), Population Census (CP) and Labour Force Survey (LFS). This simultaneity happens every ten years (last time was October 2001) and estimations can only benchmarked at that precise moment. The comparison supply-demand is done by branches of activity. The statistical unit used is the working position, or job. This includes full and part time jobs (main and secondary); the part-time jobs are transformed in FTE in a final step of the calculation. Besides the main method just explained, part-time jobs, concealed and multiple jobs in specific industries are covered with other sources: agriculture (Census of Agriculture 2000 used instead of CIS), construction (labour is estimated in relation to output observed), restaurants, transport and domestic work. Additionally there are some other special studies for marginal occupation (non-resident foreigners, students, and landlords) and secondary non-regular labour. 1 Main sources used to pre-fill this questionnaire are: For part I (persons): GNI Inventory, chapter 7 and ISTAT reply to Eurostat questionnaire in Sept For part II (hours): methodological note Le ore lavorate per la produzione del PIL una prima stima dal 1993 al 2003, dated 22 april Downloaded from website ISTAT. 2 To my knowledge, ISTAT never used the term labour accounts, but the Italian methods arguably look like a satellite labour account. 3 A word on terminology: the term comparison will be used whenever labour supply estimates are put side by side with labour demand estimates, whereas the word integration will be used whenever different sources of supply (resp. demand) are combined to improve their coverage, accuracy, etc. It is important to distinguish between these two processes of comparison and integration because they are intrinsically different.

2 The outcome of the labour estimates is number of FTEs classified by branches of activity (x101) and by kind of labour (regular/concealed). The bulk of the explained procedure is after applied at a regional level (x20). Table Main sources used to estimate labour input Source Type of information 2001 Current years Households and businesses A Population Census Number of reference form in Chap. 11 Persons in employment, resident, by place X SS47 of work B Labour Force Survey Persons in employment, resident X X SS50 C Multipurpose survey Domestic sector X SS42 D Census of Industry, Services and Institutions Registered employment, main and secondary jobs X SS48 E Agricultural Census Agricultural sector, main and secondary 2000 SS1 jobs year F Survey on the economic results of farms (REA) Working days and hours worked of selfemployed 2000 year G ASIA business register Number of workers at sectoral level X X Chp.11 H INPS data on employees and family workers Persons employed by households and businesses, quasi-subordinate workers, X X OS27 I ISTAT business accounts surveys L Balance-sheet data for specific sectors of activity M Annual survey on industrial production (Prodcom). N Administrative data and statistical surveys on particular types of persons in employment O Administrative data for particular sectors of activity P Survey on local units of enterprises Institutions Q General Accounts Department, Ministries and other authorities foreigners Number of workers X SS21 Energy, tobacco, railways, post, telephone, credit, insurance Production by products and economic activity Non-resident foreigners, Wages Guarantee Fund, part-time workers Freight and passenger transport by road, research and development, private education Workers by local units R ISTAT surveys on the public institutions Municipalities, mountain communities, provinces, regions S INPS data on private NPIs Membership organisation and other sectors X X OS3, OS11, OS year X X Chap OS27 X X OS6, OS34 X X X OS13, OS14, OS15, OS17, OS19, OS21, OS23, OS24, OS25, OS26 X OS26 X X OS27 1

3 Benchmark year estimation The benchmark 2001 estimations are divided in four main steps: 1. INTEGRATION AND HARMONISATION OF SOURCES 1.1. Harmonisation of employment concept according to the national account definitions Re-classification of Population Census and LFS data according to working place Conversion of data to annual averages. Territorial adjustments (domestic/national workforce) Adjustments to LFS data to put in line with labour concept in ESA-95. The data for the transition items are coming from Population Census, LFS, household surveys (multipurpose), Ministry of Defence, among others Integration sources The sources are integrated (separately for supply and demand of labour) in order to get the best of each one Demand side (business) The Industry Census is integrated with information from the Istat business register of active enterprises to capture mostly enterprises active during the year but not on the date of the Census. The CIS is supposed the most reliable source and discrepancies added to it. Linkage done at micro level (enterprise). This is further supplemented with the workforce obtained directly from general Government, by Agricultural Census and special branches (where concealed work is not present or relevant) captured from direct and reliable sources (energy, credit, telecommunications, railways, tobacco) Supply side (households) The new LFS on a continuous base for 2001 is integrated with the Population Census at macro data level. Additional data for Government labour is used to improve quality of LFS data for these workers (particularly to capture employment missing in the sampling LFS and to fix misclassifications of activity) The industry classification of the household data is analysed with 2 aims: (a) to correct possible reporting errors confusion of activity of employer enterprise and place of work (LKAU); (b) to reclassify household at LKAU-level. 2. COMPARISON SUPPLY SIDE-DEMAND SIDE The comparison is done as explained before, at LKAU-level. It is done by activity (x101 branches). Imbalances between observed supply and demand lead to classify labour in regular/concealed, as follows: Labour force present in both the supply and demand sources is classified as full-time registered jobs. Labour force present only in the demand sources (and not in the supply) is classified as multiple registered jobs. Labour force present only in the supply sources is classified as full-time concealed jobs. 2

4 The result is a classification of labour force (x3 kinds of jobs: legitimate, multiple legitimate, full-time concealed), by region and industry. 3. ESTIMATION OF OTHER MISSING INFORMATION Other estimates not derived of comparison supply/demand are undertaken for: 3.1 Special industries (featured by high number of part-time jobs) Agriculture: the method is basically the same as explained in point 1.2.1, but using the Agricultural Census 2000 instead of the Census of Industry and Services. Domestic work: data from social security records combined with household survey (multipurpose). 3.2 Concealed multiple jobs Focused in activities particularly susceptible to concealed activity: Construction: Estimation from expenditure side. Employees: zero by assumption. Selfemployed: comparison between expenditure surveys and average productivities from business surveys.. Hotels and restaurants and cafes: comparison between expenditure from surveys and productivities from business surveys. Compared to general labour from supply/demand calculation. Transport: number of vehicles and assumptions Other jobs Non-resident foreigners (only for workers entered in the country illegally and/or with irregular work). The collectives included here are: (a) foreigners legally in the country but not in compliance with labour law (data from businesses, Population Census, INEA, etc); (b) concealed workers traced from normal administrative sources (residence-for-employment permits, work permits, inspections of Ministry of Labour, etc); (c) other undocumented foreigners (captured only after regularisations or amnesties plus estimations based in fails in obtaining work permits and in expired permits). This latter group is only captured a posteriori and in a non-exhaustive way. Informal jobs: This name is reserved for people not declared but with some hours worked either in agriculture (typically in their own farms) or in house maintenance (plumbing, electricity, painting, etc) 4. TRANSFORMATION OF WORKING POSITIONS IN FTE Multiple jobs and part-time jobs are transformed in full-time equivalent jobs (FTE) with coefficients. They are estimated by branch and by kind of labour, as follows: Multiple jobs: a) Legitimate multiple jobs: ratio of hours spent in the second job and hours worked by those who declared in the LFS to have one job. For specific sectors (agriculture, domestic work) other methods. b) Concealed multiple jobs: ratio of hours spent in the second job and hours worked by those who declared in the LFS to have one job. Part-time jobs: c) Part-time legitimate jobs. Estimated from INPS (weekly worked hours). 3

5 d) Wages Guarantee Fund ( Cassa Integrazione Guadagni ). These are employed workers temporarily outside the production process. Source: INPS allocated in branches as in CIS. e) Part-time concealed jobs (construction, horeca services transport). From LFS derived estimates. Estimates for current years The extrapolation method is modest when compared to the extensive work in the benchmark year. However it mobilises all the available sources in current years. In lack of more frequent censuses it is impossible to overcome this problem. For years other than 2001 the estimates of labour demand and supply are extrapolated independently by means of indexes derived from current surveys. On the demand side, the main sources for enterprises and private institutions are business surveys, ASIA register and administrative data (INPS); for General Government, the main source is provided by the General Accounts Department, Ministries and other authorities.. On the supply side, the source for extrapolation is LFS. Later on the comparison supply/demand is repeated as explained for the benchmark year. Simplified versions of the remaining estimations of concealed/multiple/part-time estimations are performed every year, according to the availability of information for the period. Finally the FTE are recalculated annually too. Question 1.1: What is the main original source for employment in the national accounts (e.g. administrative source, Labour Force Survey, business survey, other)? Briefly describe this source, its coverage, its availability and whether it is in terms of jobs or persons. If necessary, please specify sources for different parts of employment (in particular if sources differ between employees and self-employed, and/or between economic branches). In this case, please provide answers to the following questions for each of these sources where necessary. Given the complexity of the labour accounts estimation method, the risk of oversimplification is high. However, for the benchmark year, in most situations the CIS has eventually prevalence over the others so it is implicitly the most reliable source. The population census is the other main source for the benchmark. For the extrapolation to other years, LFS is the main source. Question 1.2: Which calculations are made to pass from persons employed at a specific point-in-time (or several points) to annual averages? Data provided by LFS on a continue base can be expressed in annual average and on quarterly base. With respect to census data, they are transformed into annual average through coefficients, estimated using administrative data (mainly INPS) where workers are registered monthly. 4

6 Question 1.3: Which adjustments are made to pass from jobs to the concept of persons (if the original source is in terms of jobs), or vice versa? Jobs and persons employed are estimated independently. To simplify, LFS provides estimates on persons employed (the main integration to LFS data is represented by foreign non resident workers). Data provided by entrerprises and institutions (estimated from industry and services census in the benchmark year and updated as described in paragraph Estimates for current years..) represent the estimate of legitimate jobs. Indirect methods (for the current years and for the benchmark year) and the method of comparison of censuses (only in the benchmark year) are used to account also for concealed jobs. Question 1.4: Which adjustments are made in terms of coverage (See ESA to for details)? This means specifically the exclusion of residents working for non resident producer units and the inclusion of non-residents working with resident producer units. If necessary, also specify adjustments for military (including specifically conscripted forces) and other collective households not covered by the main source. Residents working for non resident producer units are estimated from Population Census in the benchmark year and updated through indicators. Non-residents working with resident producer units are implicitly covered by business surveys if legitimate. Concealed foreign workers are estimated using administrative data (see paragraph 3.3) Militaty (including specifically conscripted forces) are provided by the Ministry of Defence. Question 1.5: Which exhaustiveness adjustments are made (i.e. adjustments to take into account the unobserved economy)? Explained above Question 1.6: Which, if any, other adjustments are made (e.g. inclusion of resident workers below the age threshold considered in source statistics, prisoners at work See ESA to and to for details)? At the moment, no adjustments are made. Question 1.7: In cases where for some part of employment, Labour Force Survey data has neither been used as the main source nor for major adjustments, please explain why? Are LFS data used for cross-checking? Are differences monitored? Not applicable 5

7 II. Actual hours worked Question 2.0: Describe succinctly the architecture of your estimation method for hours worked. The Italian national account approach is based on the estimation of a per-capita of hours actually worked. That choose assures a quality analysis on micro-data in a more accurate way that an estimation of the total amount of hours directly from a source of information cannot assure. Starting from the per capita undergoing quality controls, it is also possible to use more exhaustive employment estimations as expanding coefficients. In particular, the Italian approach to estimate the total amount of hours worked consists to multiply jobs for an estimate per capita average of annual hours per different types of jobs. Employee jobs have been split in the following typologies: Registered jobs in enterprises by size Registered jobs in market NPISH Registered jobs in non market NPISH Employees in the General Government sector Registered secondary job Unregistered main jobs Unregistered secondary job Non resident and unregistered foreign workers Self-employment jobs have been split in the following typologies: Registered jobs in market enterprises (excluding quasi-subordinate workers) Registered jobs of quasi-subordinate workers Secondary job of entrepreneurs and own-account workers Secondary job of family workers Unregistered jobs The main information sources on per-capita hours actually worked are the following: LFS, business surveys (i.e. SBS), labour costs statistics. LFS is a continuous survey since 2004 (i.e. since 2004 the reference period covers all the weeks in the year). Household surveys, as opposed to businesses surveys, allow identifying hours worked in main and secondary jobs separately. Business surveys, although more precise in theory, do not give information on self-employed, they do not cover all activities (e.g. agriculture, Public administration, etc.) and they do not report labour concealed for fiscal reasons. Moreover, businesses surveyed tend to report hours paid instead of hours worked. Household surveys overcome all these problems. On their side, they tend to report hours usually worked. 6

8 All in all, business surveys offer estimates of average hours worked by employees in certain market activities of some industries (NACE C-K, M, N, O) and by employer size. On its side LFS offers data for employees and self-employed covering all 60 NACE divisions. Whenever available, business surveys information was used. Experience showed that these sources tended to report hours paid instead of hours worked. An adjustment was introduced to palliate this problem. A coefficient of adjustment has been provided by the Labour Cost survey dated 2000 ant it has been applied to the time series. The correction has been a consequence of first analysis of business surveys micro-data that have put in light a trend of small enterprises to provide data on hours paid and not on hours actually worked. This results have been confirmed by means of some telephone calls to enterprises. In order to reduce distortions, it has been decided to apply a correction procedure using the 2000 Labour Cost survey and the 2002 survey on Wages and salaries structure for the following aims: The above surveys report both paid and actually worked hours; They are conducted each 4 years in alternative way in order to have data available each two years They cover the industries C-K for enterprises with 10 workers and over (with the extension to 1-9 workers for the survey on Wages and salaries dated 2002). A sub-sample of enterprises of the labour cost survey (the all sample is around enterprises) has been analysed and an average difference by stratum of 240 hours between paid and worked hours has been estimated (around 12% of hours paid). Then the survey on the accounts of small and medium enterprises (PMI) has been analysed. The procedure of correction is based on the estimation of per capita average data for two variables, hours worked and hours paid, for the reference survey the LCS by stratum (size of enterprise and industry). Some per capita average data on hours worked and hours paid have been estimated and they have been represented the limit values for discriminate at level of industry possible outliers of PMI survey. Using specific rules, PMI data of hours actually worked have been corrected if not inside an upper and lower bound estimated starting from the LCS.. The application of procedure has led to an average decrease around the 5,2% of the original value (proximately 90 hours in absolute value). No solution has been identified at the moment for enterprise with less than 10 workers. Whenever business surveys information was not available (i.e. for self-employment and for employees of non-covered industries), LFS was used. The old non-continuous LFS had the problem of introducing distortions in the seasonal profile whenever the interview s reference week was distant from typical periods of holidays. This led to over-estimate of average hours. This problem was reduced by adjusting with information on holiday periods gathered from social surveys to households (annual multipurpose survey). This correction will be eliminated when with the revision of hours actually worked. 7

9 Hours worked in Government sector S.13 are compiled from comprehensive information from Government accounts (Ragioneria Generale dello Stato), while hours worked in the financial institutions have been gathered by a reconstructed time series of the Italian bank association (ABI) that can provide information on the 90% of the employment in the sector. Estimates of hours actually worked in the GGS have been made adopting a method by component. According to this method, the contractual number of hours are the benchmark; then the components which are not worked are deleted out and a measure of extra hours actually employed is added. Sources information permits the realization of estimates for each professional category in each collective bargaining area. The method can be summarized as follows: Haw = Hcon (Habs+Hhol)-Dpart + Hext (1) where: Haw are the hours actually worked in the GGS; Hcon denotes the total amount of hours theoretically workable considering employees as all full-timers: it is obtained with the per capita hours workable full-time by contract 4 multiplied by the annual average of persons employed; Hhol is the number of hours corresponding to the holidays occurring during the working time in the reference period; Habs represents the actual number of hours not worked because of vacations or illness, strikes, leaves (paid or not), etc.; Dpart are hours missed for part-time employment obtained as the difference between full-time hours workable and the actual hours worked by part-timers; Hext are paid overtime hours worked obtained by dividing the amount paid for overtime by the hourly rate of extra work. During the application of this approach several issues were encountered. The first regarded the attribution of the appropriate number of days workable in a week, detailed for each professional category and for each collective bargaining areas; this is because Italian contractual legislation allows a flexibility of choice between 5 or 6 workable days in a week, and the number of days workable per week influences the estimation of the amounts of absences and holidays (i.e. Habs and Hhol of notation 1) 5. The second issue encountered was the estimation of the hourly rate of the extra work done by public high management. In collective bargaining overtime for this category is not provided, but the Annual Account data show an amount paid for such activity done by public high management. The problem was resolved by applying an increase (around 10%) to the maximum hourly rate paid for overtime in each collective bargaining area. The solution of increasing the hourly rate of high management will not be applied in the next revision of hours actually worked. So, the solution that will be adopted is that one of applying the maximum hourly rate paid in each collective bargaining area for overtime of public high management. Question 2.1: What is the main original source for hours worked in the national accounts (e.g. administrative source such as official working hours, Labour Force 4 It corresponds to the weekly hours workable by contract multiplied by the number of weeks in a year (i.e. 52 weeks). It represents the hours that each employee should work without any absence during the year. 5 This problem is due to the availability of information on absences and holidays in terms of days and not in terms of hours. 8

10 Survey, other)? Briefly describe this source, its coverage and its ability to reflect the ILO definition of hours worked (see paragraphs of the SNA). Information on hours worked are collected in terms of per-capita of hours worked and not in terms of total amount of hours worked. This is due to the different coverage of the national accounts estimates on labour input respect to the surveys. Information on hours worked by employees are collected using the following sources of information. As explained above, SBS and LCI are given priority whenever available. When not available, they are complemented by LFS. Government and banks are calculated separately. In particular, per capita data are applied to different types of jobs by industries: legitimate in market enterprises, concealed in market enterprises, NPISH, General Government sector. In some cases, as the financial and insurance industry, the domestic services, agriculture and construction, information is gathered by industry only. Information on hours worked by self-employed are only provided by the Labour Force Survey. If necessary, please specify sources for different parts of the employed population (in particular if sources differ between employees and self-employed, and/or between economic branches). In this case, please provide answers to the following questions for each of these sources. See above If you do not compile national accounts employment data in terms of hours worked for the total economy, please indicate when you intend to do so. Not applicable Question 2.2: Which adjustments are made to transform the original source to adapt it to the concept of working hours as defined in the national accounts? Please specify adjustments: to take into account holidays and annual leaves, for sickness leaves, for strikes and temporary lay-offs, for paid but unreported overtime, for unpaid overtime. Per-capita hours worked derived by business survey are not adjusted for the above items; the only one adjustment considered is that one related to the differences between hours paid and hours actually worket (see Question 2). Labour force results are considered exhaustive and are adjusted only in terms of industry detail: if per capita data on hours worked of self-employed or of employees is not coherent by industry during the time, some coefficients of corrections are applied. Question 2.3: Is a specific adjustment made for unreported overtime of selfemployed? On the contrary, if LFS data is used, do you introduce a downward adjustment to reported actual hours by self-employed? 9

11 Anything specific adjustment is done for unreported overtime. All the adjustments done to the per-capita data of the labour force survey are described above. In particular, we validate LFS data at six macro-industry (according to the ESA95 classification) and when the rate of changes vary a lot the original data at 2 digit of the Nace classification is changed with the average data of macro-industry. Question 2.4: What assumption is made regarding the hours worked by persons included in the context of exhaustiveness adjustments? Comprehensiveness of the estimates is granted from the component number of jobs and number of persons. See section 1 of the questionnaire above. Question 2.5: Which, if any, other adjustments are made Adjustments are done not for exhaustiveness but only for reaching the coherence in terms of definition and for quality profile. Question 2.6: If necessary, describe any additional calculations needed to derive total hours worked and average hours worked from the sources and adjustments specified above. After having identified the better per-capita for each types of job positions, the total amount of hours worked is obtained adding the total amount of hours worked by primary registered jobs, multiple registered jobs, multiple unregistered jobs, unregistered workers.. 10

12 Annex: bridge table between LFS and national accounts This is a standard annex that may or may not fit with the sources and methods used in your country. The annex aims to reconcile the results of LFS with national accounts, both for estimates of persons (annual and quarterly) and annual hours worked. This information is most valuable to quantify the adjustments explained in sections I and II of the questionnaire. If you use LFS for the national accounts employment estimates, please quantify the adjustments in place for the periods indicated below (the figures included here are the latest estimates available at Eurostat). Complete at least one column for one of the periods; the information for several columns is welcome. If you do not use LFS for these employment estimates, but you use it for control or verification, please give your interpretation of the figures. Persons employed (annual data) LFS 6 IV quarter Employment residents, by place of work Employment in extra- territorial organisations Integration done ,771 n.a. n.a. n.a in 1991 comparing population census and LFS at a microlevel Employment living in institutions Conversion to annual average Conscript force 104 Non residents and unregistered foreigners workers The annual LFS results indicated here are actually the average of the quarters in the year. 11

13 Employment adding to LFS data by industry National accounts (domestic concept) 474,1 23,393,1 Persons employed (quarterly data) (the following data will be revised) 2004Q1 2004Q2 2004Q3 2004Q4 LFS 22,065 22,438 22,485 22,630 (Insert here rows for your adjustments) National accounts (domestic concept) 24,371 24,583 24,560 24,470 Annual hours worked (the following data will be revised ) LFS 41,465,160 40,745,861 40,986,680 39,772,233 (Insert here rows for your adjustments) National accounts (domestic concept) 43,641,177 43,876,727 43,959,308 NA 12